Code documentation for paper "Spatially Explicit Action Research for Coastal Fisheries Management"

Summary
Initialise
Run the non-parametric statistics
SPATIAL ANALYSIS
Generate plots
- generate histogram of distance from shoreline
Mapping
- Get the basemaps and panels
GIS data processing using GRASS and R
Generate interactive maps

Summary

This is supplementary material to allow interested persons to replicate the results of the data. This is packaged along with the code and the dataset. The code blocks embedded within this document has been extracted (tangled) into the file FisheriesMappingScript.R. You need to be connected to the internet for the code to work as the OSM maps are yanked off the web. The script was created using Emacs 2.5 with ESS and org-babel.

Details of the system and R version used are:

R.Version()

platform	arch	os	system	status	major	minor	year	month	day	svn.rev	language	version.string	nickname
x86_64-pc-linux-gnu	x86_64	linux-gnu	x86_64, linux-gnu		3	4.4	2018	3	15	74408	R	R version 3.4.4 (2018-03-15)	Someone to Lean On

Initialise

This will set up the environment, unzip and create the data folder and load the required libraries.

NOTE: You should be running this script from within the folder you intend to copy the data to.

Set up the environment

setwd("./")
print(paste("The working directory is:", getwd()))

[1] "The working directory is: /home/udumbu/rsb/Papers/Fisheries/Mapping/PLOS/rscript"

Load the libraries

In case the output here throws an error, you have a library missing or, the library is outdated. Please use the command install.package("NameOfPackage", dependencies=TRUE) to install or update the missing packages.

library(tools)
library(spatstat) # spatial point pattern analysis
library(spdep) # spatial statistics
library(raster) # raster operations
library(rgeos) # import export of geographical data
library(ggplot2) # graphing and visualising
library(ggmap) # mapping in ggplot
library(ggsn) # map furniture
library(sp)
library(grid) # manipulate ggplot2 objects
library(gtable)# manipulate ggplot2 objects
library(gridExtra) # graphing and visualising
library(ggExtra) # graphing and visualising
library(lattice) # graphing and visualising
library(maptools) # tools to manipulate geographical data
library(rgdal) # manipulate raster images
library(rgl) # 3-D visualisation
library(lmtest)
library(pgirmess) # kruskal wallis multi component analysis; load pgirmess library
library(extrafont) #load fonts for Plos One
library(rgrass7)
library(gstat)
library(doBy) # summary stats
library(dunn.test) #post hoc tests after Kruskal Wallis
library(plotly) # generate interactive maps

x
plotly
dunn.test
doBy
gstat
rgrass7
XML
extrafont
pgirmess
lmtest
zoo
rgl
rgdal
maptools
lattice
ggExtra
gridExtra
gtable
grid
ggsn
ggmap
ggplot2
rgeos
raster
spdep
spData
Matrix
sp
spatstat
rpart
nlme
spatstat.data
tools
stats
graphics
grDevices
utils
datasets
methods
base

Datasets

Check to see if datasets exist, else unzip the data file. Then create a folder to hold the results. Note the datasets include a data directory with the ESRI shapefiles and a GRASS database.

if(!dir.exists("./Data"))unzip("Datasets.zip", exdir = "./Data")
if(!dir.exists("./grassdata"))unzip("grassdata.zip", exdir = "./grassdata")
dir.create("./Results")

Warning message:
In dir.create("./Results") : './Results' already exists

Read in the data

fc.data <- readOGR(dsn = "./Data", "BtDistUTM", integer64 = "warn.loss", stringsAsFactors = FALSE) # data is in UTM 44North EPSG 32644
fc.data$mesh1mm <- as.numeric(fc.data$mesh1mm) ## NAs are for line and hook
fdf <- as.data.frame(fc.data) # ggplot only works on dataframes
spatial.df<-read.csv("./Data/spatial2017.csv", stringsAsFactors=FALSE) # cleaned data for npstat.R

OGR data source with driver: ESRI Shapefile 
Source: "/home/udumbu/rsb/Papers/Fisheries/Mapping/PLOS/rscript/Data", layer: "BtDistUTM"
with 3427 features
It has 26 fields
Integer64 fields read as signed 32-bit integers:  ID slno crew_size no_boats1 no_boats2 tot_boats boat_lengt survey_mon no_days
Warning message:
NAs introduced by coercion

Functions and initialisation scripts

Some functions to assist with the plotting. Note that code has been borrowed from various sources, mostly stackexchange.

Extract legend

Shares a legend between two ggplot graphs. Source: https://github.com/hadley/ggplot2/wiki/Share-a-legend-between-two-ggplot2-graphs

g_legend<-function(a.gplot){
tmp <- ggplot_gtable(ggplot_build(a.gplot))
leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
legend <- tmp$grobs[[leg]]
return(legend)
}

Write both eps and tiff

eps.tiff <- function(plot.cmd, flnm, width.in, height.in){
  tiff(filename = paste0(flnm, ".tiff"),
    width = width.in, height = height.in,
    units = "in", compression = "lzw", pointsize = 12, res = 450, family = "Arial",
    type = "cairo", antialias="subpixel")
  plot(plot.cmd)
  dev.off()
  cairo_ps(filename = paste0(flnm, ".eps"),
    width = width.in, height = height.in, pointsize = 12,
    onefile = FALSE, family = "Arial", bg = "white",
    antialias = "subpixel", fallback_resolution =450)
  plot(plot.cmd)
dev.off()
}

Create weight files for tests of auto-correlation

get.weights <- function(long, lat){
    xy <- cbind(long, lat)
    map_nb <<- knn2nb(knearneigh(xy, longlat=TRUE))
    map_nb <<- make.sym.nb(map_nb)
    map_w <<- nb2listw(map_nb,glist=NULL,style="W",zero.policy=FALSE)
    map_b <<- nb2listw(map_nb, glist=NULL, style="B", zero.policy=FALSE)
}

Run the local Geary's test and output the data in a formatted table

gry.test <- function(x){
  df <- as.data.frame(matrix(nrow=1, ncol=5))
  names(df) <- c("data.name", "Geary C statistic", "Expectation", "Variance", "p.value")
  gt <- geary.test(x, map_w, randomisation=TRUE, zero.policy=NULL, alternative="greater", spChk=NULL, adjust.n=TRUE)
  df[1,1] <- gt$data.name
  df[1,2:4] <- gt$estimate
  df[1,5] <- gt$p.value
  write.table(df,"./Results/GearyTest2.csv", append=TRUE, row.names=TRUE)
  return(df)
}

Initialise GRASS

Ensure you're pointing this to the correct folder (check your installed GRASS version) do this only when running the script outside GRASS. Else you need to run R from within GRASS. Type ?rgrass7 for more information.

initGRASS(gisBase = "/usr/lib/grass74/", home = tempdir(), 
  gisDbase = "./grassdata",
  location = "fisheries", mapset = "craftdist", override = TRUE)

gisdbase    ./grassdata 
location    fisheries 
mapset      craftdist 
rows        162 
columns     111 
north       1407232 
south       1245267 
west        365066.1 
east        476022 
nsres       999.7841 
ewres       999.603 
projection  +proj=utm +no_defs +zone=44 +a=6378137 +rf=298.257223563
+towgs84=0.000,0.000,0.000 +to_meter=1

Plot Panels

plot.panels <- function(panels){
craft.map <- ggmap(pdymap.bw,
                   base_layer = ggplot(aes(x = long, y = lat), data = fdf))  +
    geom_point(shape = 1) +
    labs(x = 'Longitude', y = 'Latitude') +
    facet_wrap(as.formula(paste("~", panels)))
    ggsave(filename = paste0(panels, ".png"))
    }

Multi-plot panels

qmplot.panels <- function(panels){
craft.map <- qmplot(long, lat, data = fdf, maptype = "toner-lite", extent = "panel")  +
    geom_point(shape = 1) +
    theme(plot.caption = element_text(colour="black", size=6, face="bold"),
    axis.text = element_text(colour="black", size=7)) +
    labs(x = 'Longitude', y = 'Latitude', caption = "NOT TO SCALE") +
    facet_wrap(as.formula(paste("~", panels)))
ggsave(filename = paste0("./Results/PlotPanel", panels, ".png"))
print(craft.map)
}

Create new GRASS layers based on a list of input files through sub-setting
Creates new layers based on input list by subsetting maps through v.extract. Use as lapply(X = map.list, FUN = subset.map, colhd). Here in.list is the filter criteria for extracting data; colhd is the field in which these filter criteria lie. For example for the colhd = fishery, the list is ("trawl", "gillnet"….)
```
subset.map <- function(map.list, colhd){
map.out <- gsub(" ", "", map.list, fixed = TRUE) #remove whitespace
map.out <- paste0(map.out,"_",colhd) # append col name to output map
execGRASS("v.extract", input="BtDistUTM@PERMANENT", type="point",
          where = paste0(colhd, " = \'", map.list,"\'"),
          output=map.out, flags = "overwrite", intern = TRUE)
          }
```

Subset mesh size

Input is size range as lwr (lower boundary) upr (upper boundary) and nm (name) of output. Use as mapply(FUN, lwr=, upr=, nm=)

subsetmesh <- function(lwr, upr, nm){
query <- paste("meshsize >=", lwr, "and meshsize <", upr)
execGRASS("v.extract", input="BtDistUTM@PERMANENT",
type="point", where=query,
          output=nm, flags = "overwrite")
          }

Generate quadrat

The function generates a quadrat of 1km, sum number of boats and puts the data to the centroid of that quadrat for subsequent visualisation. Use with mapply with x = net.maps, y = net.ptmaps.

build.quads <- function(x, y){
execGRASS("v.select", parameters = list (ainput="onekmgrid@craftdist", atype="area", binput=x, btype=c("point","area"), output="quad1k@craftdist", operator="contains"), flags = "overwrite") # select only those grids with boats
## see if a table exists
if(identical(execGRASS("v.db.connect", flags = "c", map = x, intern=TRUE), character(0)))
execGRASS("v.db.addtable", map=x, columns="tot_boats INT") # create table and add col
## execGRASS("v.db.addcolumn", map=x, columns="tot_boats INT") #add column if table exists
if(execGRASS("v.info", map=x, flags = "t", intern = TRUE)[2] != "points=0"){
execGRASS("v.vect.stats", parameters = list(points=x, areas="quad1k@craftdist", count_column="count", method="sum", points_column="tot_boats", stats_column="sum_bts"))
## count the points in each grid
    execGRASS("v.extract", input="quad1k@craftdist", type="centroid", output="quad1kcent@craftdist", flags = "overwrite")
    execGRASS("v.type", input="quad1kcent@craftdist", output=y, from_type="centroid", to_type="point", flags = "overwrite") # generate a centroid map
    }else{
    print("layer is empty")
    }
    }

Create a data frame of IDW interpolated predictors

Create a data frame of IDW interpolated predictors for each of the fisheries and rbind into a single data frame for plotting. Use with mapply and rbind with x = net.ptmaps, y = fishery.maps

run.idw <- function(x, y){
boatdens <- readVECT(x, plugin=NULL)
bt.ll <- spTransform(boatdens,  CRS("+proj=longlat +datum=WGS84")) # change crs to lat long
## bt.ll$efrt <- bt.ll$sum_bts
## bt.pp <- as.ppp(bt.ll)
grd <- as.data.frame(spsample(bt.ll, "regular", n=100000))
names(grd) <- c("x", "y")
coordinates(grd) <- c("x", "y")
gridded(grd) <- TRUE  # Create SpatialPixel object
fullgrid(grd) <- TRUE  # Create SpatialGrid object
proj4string(grd) <- CRS(proj4string(bt.ll)) 
## Interpolate the surface using a power value of 2 (idp=2.0)
dat.idw <- gstat::idw(sum_bts~1,bt.ll,newdata=grd,idp=2.0)
idw.df <- as.data.frame(dat.idw)
idw.df$facet <- y
return(idw.df)
}

Plot the quadrat count map

Function to ggplot the map. Input is ggdat created from the call above.

plot.quadcount <- function(plot.data){
ggdat <<- plot.data
plot <- ggmap(pdymap.bw, base_layer = ggplot(aes(x = x, y = y, z = var1.pred), data=ggdat)) +
facet_wrap( ~ facet, ncol = 3) +
geom_raster(aes(alpha=var1.pred))+    
    labs(alpha = "Boats/km2", x = "long", y = "lat") +
    geom_contour(alpha = .25) +
    coord_equal()
    return(plot)
    }

Multi-plot the quadrat count map

qmplot.quadcount <- function(ggdat){
plot <- qmplot(x = x, y = y, data = ggdat, maptype = "toner-lite", extent = "panel", alpha = I(.0)) +
    theme(plot.caption = element_text(colour="black", size=6, face="bold")) +
    theme(axis.text = element_text(colour="black", size=7)) +
    labs(alpha = "Boats/km2", x = 'Longitude', y = 'Latitude', caption = "NOT TO SCALE") +
    facet_wrap( ~ facet) +
    geom_raster(aes(alpha=var1.pred))+
    geom_contour(alpha = .25, aes(x = x, y = y, z = var1.pred)) +
    coord_equal()
    return(plot)
    }

Read a list of vector files

read.vect.list <- function(x, facet.x){
vect.x <- readVECT(x)
df.x <- as.data.frame(vect.x)    
df.x$facet <- facet.x
return(df.x)
}

Multi-plot a 2-d kernel density dataset

qmplot.dens2d.mesh <- function(ggdat, fig.filename){
out <- by(data = ggdat, INDICES = ggdat$facet, FUN = function(m) {
    m <- droplevels(m)
    qmplot(x = long, y = lat, data = ggdat, maptype = "toner-lite", extent = "panel", zoom=10, geom="text", label="") + # force extents to fit data locations #qmplot
        geom_point(data = m, aes(x = long, y = lat, colour=meshsize)) +
        stat_density2d(data = m, aes(x = long, y = lat)) +
        labs(
            x = NULL,
            y = NULL ,
            title = m$facet[1],
            colour = "Mesh \nSize (mm)",
            caption = NULL # "Not to scale."
        ) +
        coord_equal()
})
grb.out <- do.call(arrangeGrob, c(out, list(ncol=3, padding = 0)))
x.lab <- textGrob("Longitude", gp=gpar(fontsize=14))
y.lab <- textGrob("Latitude", rot=90,gp=gpar(fontsize=14))
grid.arrange(grb.out, left=y.lab, bottom=x.lab, widths=c(100, 0.5, 0.0))

if(file_ext(fig.filename)=="tiff"){
    ## fig.file <-
    tiff(filename = fig.filename, width = 12, height = 9,
         units = "in", compression = "lzw", pointsize = 12, res = 300, family = "Arial",
         type = "cairo", antialias="subpixel")
    grid.arrange(grb.out, left=y.lab, bottom=x.lab, widths=c(100, 0.5, 0.0))
    dev.off()
}else{
    cairo_ps(filename = fig.filename, width = 12, height = 9, pointsize = 12,
             onefile = FALSE, family = "Arial", bg = "white",
             antialias = "subpixel", fallback_resolution =600)
    grid.arrange(grb.out, left=y.lab, bottom=x.lab, widths=c(100, 0.5, 0.0))
    dev.off()
  }
}

Run the non-parametric statistics

Summarise the data

Summary of mesh sizes by type of net

summaryBy(mesh1mm~fishery,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

    fishery mesh1mm.median mesh1mm.sd mesh1mm.mean
1   Gillnet             46  16.642336    46.281911
2      Lift             12   0.000000    12.000000
3      Line              6   3.953912     9.115942
4 Ringseine             28  37.736702    51.384977
5  Scoopnet             12   0.860663    12.074074
6     Trawl             20   1.387490    20.064257

Summary of depth of fishing by type of net

summaryBy(depth_m~fishery,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

    fishery depth_m.median depth_m.sd depth_m.mean
1   Gillnet          12.25   5.188013    13.397689
2      Lift          14.08   2.485833    14.355000
3      Line          24.51  21.624134    28.640435
4 Ringseine          24.69  21.499570    30.607786
5  Scoopnet           8.05   2.370998     7.828963
6     Trawl          24.51  15.991006    27.336233

Summary of distance to shore by net

summaryBy(shoredist_km~fishery,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

    fishery shoredist_km.median shoredist_km.sd shoredist_km.mean
1   Gillnet                1.82        2.430419          2.697331
2      Lift                4.11        1.611076          3.955000
3      Line               13.26       11.214906         12.904348
4 Ringseine               10.30        8.268917         10.737324
5  Scoopnet                0.80        1.521430          1.074370
6     Trawl               10.88        8.450241         10.823719

Summary of mesh sizes by craft type

summaryBy(mesh1mm~boat_type1,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

  boat_type1 mesh1mm.median mesh1mm.sd mesh1mm.mean
1        frp             52  20.714772     41.03734
2    frp_big             75  25.859127     76.63889
3 kattumaram             44  12.742809     43.31197
4      trawl             20   2.627919     19.57377
5     vallam             34  38.841742     53.41304

Summary of depth of fishing by craft type

summaryBy(depth_m~boat_type1,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

  boat_type1 depth_m.median depth_m.sd depth_m.mean
1        frp          13.72   5.674216     14.40787
2    frp_big          16.09   4.338676     16.57528
3 kattumaram           9.69   2.802077     10.15599
4      trawl          25.24  16.173801     28.72341
5     vallam          26.70  21.598749     33.53738

Summary of distance to shore by craft type

summaryBy(shoredist_km~boat_type1,data = spatial.df, FUN=c(median, sd, mean), na.rm=TRUE)

  boat_type1 shoredist_km.median shoredist_km.sd shoredist_km.mean
1        frp               2.585        2.644221          3.288047
2    frp_big               3.975        1.677750          4.062222
3 kattumaram               0.880        1.063816          1.213758
4      trawl              11.355        8.509531         11.660279
5     vallam              11.450        8.086057         12.068750

Plot to see the distribution of the data

Overall distribution via frequency curves/histograms and scatterplots

p <- ggplot(spatial.df, aes(depth_m, shoredist_km)) + geom_point() + theme_classic()
ggExtra::ggMarginal(p, type = "histogram")

Data are strongly positive skewed

Plotting variables to visualise skew

Shoreline distance

ggplot(spatial.df, aes(shoredist_km)) + geom_freqpoly(bins = 20)

Depth

ggplot(spatial.df, aes(depth_m)) + geom_histogram(bins = 20)

Mesh size
```
ggplot(spatial.df, aes(mesh1mm)) + geom_freqpoly(bins = 10)
```
Data is non-normal, strong positive skew, we need to use non-parametric tests.

Distribution of fishing points by distance and bottom depth

ggplot(spatial.df, aes(shoredist_km, depth_m)) + geom_point()

Plotting box plots to show differences between fisheries and craft types

Plotting for bottom depth by fishery (net type)

ggplot(spatial.df,aes(fishery, depth_m)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE) #, outlier.shape=NA)
p<-ggplot(spatial.df,aes(fishery, shoredist_km)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 38))
b<- p + xlab("Fishery") + ylab("Distance to shore (km)") + ggtitle("Distance to Shore") 
c<-b + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
d<-c + theme(axis.text = element_text(size=10, face="plain"))
e<-d + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
e

Plotting for bottom depth by fishery (net type)

f<-ggplot(spatial.df,aes(fishery, depth_m)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 70))
g<- f + xlab("Fishery") + ylab("Bottom depth (m)") + ggtitle("Depth of Operation") 
h<-g + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
i<-h + theme(axis.text = element_text(size=10, face="plain"))
j<-i + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
j

Plots are very similar suggesting the obvious correlation between depth and distance from shore. Put the two plots together and save as eps and tiff.

plot.cmd <- grid.arrange(e, j,  layout_matrix = matrix(c(1,2), ncol=2, byrow=TRUE))
flnm <- "./Results/FishingDistancesDepth_byGear"
eps.tiff(plot.cmd, flnm, width.in = 12, height.in = 6)
plot.cmd

Plotting for bottom depth and distance of fishing by craft type

ggplot(spatial.df,aes(boat_type1, depth_m)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE)  #, outlier.shape=NA)
p<-ggplot(spatial.df,aes(boat_type1, shoredist_km)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 35))
p
b<- p + xlab("Type of fishing craft") + ylab("Distance to shore (km)") + ggtitle("Distance to Shore") + scale_x_discrete(breaks=c("frp", "frp_big", "kattumaram","trawl","vallam"), labels=c("Small FRP", "Big FRP", "Kattumaram", "Trawler", "Vallam"))
c<-b + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
d<-c + theme(axis.text = element_text(size=10, face="plain"))
e<-d + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
e

f<-ggplot(spatial.df,aes(boat_type1, depth_m)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 65))
f
g<- f + xlab("Type of fishing craft") + ylab("Bottom depth (m)") + ggtitle("Depth of Operation") + scale_x_discrete(breaks=c("frp", "frp_big", "kattumaram","trawl","vallam"), labels=c("Small FRP", "Big FRP", "Kattumaram", "Trawler", "Vallam"))
h<-g + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
i<-h + theme(axis.text = element_text(size=10, face="plain"))
j<-i + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
j

plot.cmd <- grid.arrange(e, j, layout_matrix = matrix(c(1,2), ncol=2, byrow=TRUE))
flnm <- "./Results/FishingDistancesDepth_byCraft"
eps.tiff(plot.cmd, flnm, width.in = 12, height.in = 6)
plot.cmd

Plotting for mesh sizes across fisheries and craft type

ggplot(spatial.df,aes(boat_type1, mesh1mm)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE)  #, outlier.shape=NA)
p<-ggplot(spatial.df,aes(fishery, mesh1mm)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 115))
b<- p + xlab("Fishery") + ylab("Mesh size (mm)") + ggtitle("Fishery") 
c<-b + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
d<-c + theme(axis.text = element_text(size=10, face="plain"))
e<-d + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
e

f<-ggplot(spatial.df,aes(boat_type1,  mesh1mm)) + stat_boxplot(geom ='errorbar') + geom_boxplot(notch = TRUE, coef = 1.5, outlier.shape=NA) + scale_y_continuous(limits = c(0, 115))
f
g<- f + xlab("Type of fishing craft") + ylab("Mesh size (mm)") + ggtitle("Craft") + scale_x_discrete(breaks=c("frp", "frp_big", "kattumaram","trawl","vallam"), labels=c("Small FRP", "Big FRP", "Kattumaram", "Trawler", "Vallam"))
h<-g + theme(axis.title.x = element_blank(), axis.title.y = element_text(face="plain", size=12))
i<-h + theme(axis.text = element_text(size=10, face="plain"))
j<-i + theme(plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
j

plot.cmd <- grid.arrange(e, j, layout_matrix = matrix(c(1,2), ncol=2, byrow=TRUE))
flnm <- "./Results/MeshSizesAcrossFisheriesAndCraft"
eps.tiff(plot.cmd, flnm, width.in = 12, height.in = 6)
plot.cmd

Kruskal Wallis for non parametric

Differences in between gear, craft & regions

Fishery and depth of fishing

spatial.df$fishery.f <- as.factor(spatial.df$fishery)
kruskal.test(depth_m ~ fishery.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  depth_m by fishery.f
Kruskal-Wallis chi-squared = 1111.5, df = 5, p-value < 2.2e-16

Fishery and distance to shore

kruskal.test(shoredist_km ~ fishery.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  shoredist_km by fishery.f
Kruskal-Wallis chi-squared = 1221, df = 5, p-value < 2.2e-16

Fishery and mesh size

kruskal.test(mesh1mm ~ fishery.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  mesh1mm by fishery.f
Kruskal-Wallis chi-squared = 1756.1, df = 5, p-value < 2.2e-16

Depth and type of boat

spatial.df$boat_type1.f <- as.factor(spatial.df$boat_type1)
kruskal.test(depth_m ~ boat_type1.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  depth_m by boat_type1.f
Kruskal-Wallis chi-squared = 1477.9, df = 4, p-value < 2.2e-16

Distance to shore and type of boat

kruskal.test(shoredist_km ~ boat_type1.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  shoredist_km by boat_type1.f
Kruskal-Wallis chi-squared = 1713.8, df = 4, p-value < 2.2e-16

Mesh size and type of boat

kruskal.test(mesh1mm ~ boat_type1.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  mesh1mm by boat_type1.f
Kruskal-Wallis chi-squared = 953.14, df = 4, p-value < 2.2e-16

Differences in fishing between regions

spatial.df$region.f <- as.factor(spatial.df$region)
kruskal.test(depth_m ~ region.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  depth_m by region.f
Kruskal-Wallis chi-squared = 300.28, df = 1, p-value < 2.2e-16

Difference in distance to shore and regions

kruskal.test(shoredist_km ~ region.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  shoredist_km by region.f
Kruskal-Wallis chi-squared = 193.04, df = 1, p-value < 2.2e-16

Difference in size of mesh used and regions

kruskal.test(mesh1mm ~ region.f, data = spatial.df)

   Kruskal-Wallis rank sum test

data:  mesh1mm by region.f
Kruskal-Wallis chi-squared = 321.27, df = 1, p-value < 2.2e-16

post hoc with FDR correction - Benjamini-Hochberg adjustment

By fishery

Distance to shore

dunn.test (spatial.df$shoredist_km, g=spatial.df$fishery, method="bh", kw=TRUE, label=TRUE, 
wrap=TRUE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 1221.029, df = 5, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |    Gillnet       Lift       Line   Ringsein   Scoopnet
---------+-------------------------------------------------------
    Lift |  -1.142964
     |     0.1898
     |
    Line |  -8.697724  -0.963016
     |    0.0000*     0.2097
     |
Ringsein |  -16.06921  -1.176209  -0.709885
     |    0.0000*     0.1996     0.2756
     |
Scoopnet |   7.059364   2.370267   11.47283   16.32725
     |    0.0000*    0.0167*    0.0000*    0.0000*
     |
   Trawl |  -30.49773  -1.112491  -0.500257   0.491747  -19.42008
     |    0.0000*     0.1813     0.3305     0.3114    0.0000*


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
Gillnet - Lift       : -1.142964 (0.1898)
Gillnet - Line       : -8.697724 (0.0000)*
Lift - Line          : -0.963016 (0.2097)
Gillnet - Ringseine  : -16.06921 (0.0000)*
Lift - Ringseine     : -1.176209 (0.1996)
Line - Ringseine     : -0.709885 (0.2756)
Gillnet - Scoopnet   :  7.059364 (0.0000)*
Lift - Scoopnet      :  2.370267 (0.0167)*
Line - Scoopnet      :  11.47283 (0.0000)*
Ringseine - Scoopnet :  16.32725 (0.0000)*
Gillnet - Trawl      : -30.49773 (0.0000)*
Lift - Trawl         : -1.112491 (0.1813)
Line - Trawl         : -0.500257 (0.3305)
Ringseine - Trawl    :  0.491747 (0.3114)
Scoopnet - Trawl     : -19.42008 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

Depth

dunn.test (spatial.df$depth_m, g=spatial.df$fishery, method="bh", kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 1111.5321, df = 5, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |    Gillnet       Lift       Line   Ringsein   Scoopnet
---------+-------------------------------------------------------
    Lift |  -0.580412
     |     0.2808
     |
    Line |  -7.436650  -1.209637
     |    0.0000*     0.1415
     |
Ringsein |  -14.98615  -1.578500  -1.259924
     |    0.0000*     0.0954     0.1416
     |
Scoopnet |   10.02170   2.336669   12.21471   18.01781
     |    0.0000*    0.0182*    0.0000*    0.0000*
     |
   Trawl |  -27.68484  -1.466803  -0.909434   0.836801  -21.19019
     |    0.0000*     0.1068     0.2095     0.2157    0.0000*


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
Gillnet - Lift       : -0.580412 (0.2808)
Gillnet - Line       : -7.436650 (0.0000)*
Lift - Line          : -1.209637 (0.1415)
Gillnet - Ringseine  : -14.98615 (0.0000)*
Lift - Ringseine     : -1.578500 (0.0954)
Line - Ringseine     : -1.259924 (0.1416)
Gillnet - Scoopnet   :  10.02170 (0.0000)*
Lift - Scoopnet      :  2.336669 (0.0182)*
Line - Scoopnet      :  12.21471 (0.0000)*
Ringseine - Scoopnet :  18.01781 (0.0000)*
Gillnet - Trawl      : -27.68484 (0.0000)*
Lift - Trawl         : -1.466803 (0.1068)
Line - Trawl         : -0.909434 (0.2095)
Ringseine - Trawl    :  0.836801 (0.2157)
Scoopnet - Trawl     : -21.19019 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

Mesh size

dunn.test (spatial.df$mesh1mm, g=spatial.df$fishery, method="bh", kw=TRUE, label=TRUE, 
 wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 1756.0819, df = 5, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |    Gillnet       Lift       Line   Ringsein   Scoopnet
---------+-------------------------------------------------------
    Lift |   4.524593
     |    0.0000*
     |
    Line |   18.77161   0.075618
     |    0.0000*     0.5034
     |
Ringsein |   3.418048  -3.996420  -14.84107
     |    0.0004*    0.0000*    0.0000*
     |
Scoopnet |   25.21785  -0.024967  -0.348390   18.21847
     |    0.0000*     0.4900     0.4197    0.0000*
     |
   Trawl |   33.59573  -2.037622  -8.565356   13.43836  -11.12185
     |    0.0000*     0.0260    0.0000*    0.0000*    0.0000*


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
Gillnet - Lift       :  4.524593 (0.0000)*
Gillnet - Line       :  18.77161 (0.0000)*
Lift - Line          :  0.075618 (0.5034)
Gillnet - Ringseine  :  3.418048 (0.0004)*
Lift - Ringseine     : -3.996420 (0.0000)*
Line - Ringseine     : -14.84107 (0.0000)*
Gillnet - Scoopnet   :  25.21785 (0.0000)*
Lift - Scoopnet      : -0.024967 (0.4900)
Line - Scoopnet      : -0.348390 (0.4197)
Ringseine - Scoopnet :  18.21847 (0.0000)*
Gillnet - Trawl      :  33.59573 (0.0000)*
Lift - Trawl         : -2.037622 (0.0260)
Line - Trawl         : -8.565356 (0.0000)*
Ringseine - Trawl    :  13.43836 (0.0000)*
Scoopnet - Trawl     : -11.12185 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

By craft

Distance to shore

dunn.test (spatial.df$shoredist_km, g=spatial.df$boat_type1, method="bh", kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 1713.7937, df = 4, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |        frp    frp_big   kattumar      trawl
---------+--------------------------------------------
 frp_big |  -2.102112
     |    0.0197*
     |
kattumar |   15.37248   6.250093
     |    0.0000*    0.0000*
     |
   trawl |  -26.07743  -4.124113  -38.05761
     |    0.0000*    0.0000*    0.0000*
     |
  vallam |  -14.92838  -4.523127  -22.98987  -1.604463
     |    0.0000*    0.0000*    0.0000*     0.0543


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
frp - frp_big        : -2.102112 (0.0197)*
frp - kattumaram     :  15.37248 (0.0000)*
frp_big - kattumaram :  6.250093 (0.0000)*
frp - trawl          : -26.07743 (0.0000)*
frp_big - trawl      : -4.124113 (0.0000)*
kattumaram - trawl   : -38.05761 (0.0000)*
frp - vallam         : -14.92838 (0.0000)*
frp_big - vallam     : -4.523127 (0.0000)*
kattumaram - vallam  : -22.98987 (0.0000)*
trawl - vallam       : -1.604463 (0.0543)

alpha = 0.05
Reject Ho if p <= alpha/2

Depth

dunn.test (spatial.df$depth_m, g=spatial.df$boat_type1, method="bh", kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 1477.8973, df = 4, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |        frp    frp_big   kattumar      trawl
---------+--------------------------------------------
 frp_big |  -2.157017
     |    0.0172*
     |
kattumar |   13.59512   5.822747
     |    0.0000*    0.0000*
     |
   trawl |  -24.57939  -3.711565  -34.97901
     |    0.0000*    0.0001*    0.0000*
     |
  vallam |  -14.58273  -4.322306  -21.65827  -2.023843
     |    0.0000*    0.0000*    0.0000*    0.0215*


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
frp - frp_big        : -2.157017 (0.0172)*
frp - kattumaram     :  13.59512 (0.0000)*
frp_big - kattumaram :  5.822747 (0.0000)*
frp - trawl          : -24.57939 (0.0000)*
frp_big - trawl      : -3.711565 (0.0001)*
kattumaram - trawl   : -34.97901 (0.0000)*
frp - vallam         : -14.58273 (0.0000)*
frp_big - vallam     : -4.322306 (0.0000)*
kattumaram - vallam  : -21.65827 (0.0000)*
trawl - vallam       : -2.023843 (0.0215)*

alpha = 0.05
Reject Ho if p <= alpha/2

Mesh size

dunn.test (spatial.df$mesh1mm, g=spatial.df$boat_type1, method="bh", kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 953.1441, df = 4, p-value = 0


               Comparison of x by group                            
                 (Benjamini-Hochberg)                              
Col Mean-|
Row Mean |        frp    frp_big   kattumar      trawl
---------+--------------------------------------------
 frp_big |  -6.844527
     |    0.0000*
     |
kattumar |  -6.696647   4.966940
     |    0.0000*    0.0000*
     |
   trawl |   22.50522   12.21877   26.29041
     |    0.0000*    0.0000*    0.0000*
     |
  vallam |  -2.767481   5.150123   1.108463  -14.25951
     |    0.0031*    0.0000*     0.1338    0.0000*


List of pairwise comparisons: Z statistic (adjusted p-value)
------------------------------------------
frp - frp_big        : -6.844527 (0.0000)*
frp - kattumaram     : -6.696647 (0.0000)*
frp_big - kattumaram :  4.966940 (0.0000)*
frp - trawl          :  22.50522 (0.0000)*
frp_big - trawl      :  12.21877 (0.0000)*
kattumaram - trawl   :  26.29041 (0.0000)*
frp - vallam         : -2.767481 (0.0031)*
frp_big - vallam     :  5.150123 (0.0000)*
kattumaram - vallam  :  1.108463 (0.1338)
trawl - vallam       : -14.25951 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

By region

Distance to shore

dunn.test (spatial.df$shoredist_km, g=spatial.df$region,  kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 193.0353, df = 1, p-value = 0


               Comparison of x by group                            
                (No adjustment)                                
Col Mean-|
Row Mean |   Pondiche
---------+-----------
Tamil Na |   13.89371
     |    0.0000*


List of pairwise comparisons: Z statistic (p-value)
----------------------------------------------
Pondicherry - Tamil Nadu :  13.89371 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

Depth

dunn.test (spatial.df$depth_m, g=spatial.df$region,  kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 300.2785, df = 1, p-value = 0


               Comparison of x by group                            
                (No adjustment)                                
Col Mean-|
Row Mean |   Pondiche
---------+-----------
Tamil Na |   17.32854
     |    0.0000*


List of pairwise comparisons: Z statistic (p-value)
----------------------------------------------
Pondicherry - Tamil Nadu :  17.32854 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

Mesh size

dunn.test (spatial.df$mesh1mm, g=spatial.df$region,  kw=TRUE, label=TRUE, 
wrap=FALSE, table=TRUE, list=TRUE, rmc=FALSE, alpha=0.05)

  Kruskal-Wallis rank sum test

data: x and group
Kruskal-Wallis chi-squared = 321.2731, df = 1, p-value = 0


               Comparison of x by group                            
                (No adjustment)                                
Col Mean-|
Row Mean |   Pondiche
---------+-----------
Tamil Na |  -17.92409
     |    0.0000*


List of pairwise comparisons: Z statistic (p-value)
----------------------------------------------
Pondicherry - Tamil Nadu : -17.92409 (0.0000)*

alpha = 0.05
Reject Ho if p <= alpha/2

SPATIAL ANALYSIS

Spatial patterns

Prepare the data set for neighbourhood analysis

Create a nearest neighbour file so that each feature is assessed within the spatial context of a fixed number of its closest neighbours.

xy<-cbind(spatial.df$long, spatial.df$lat)
colnames(xy) <- c("LONG","LAT")
map_nb <-knn2nb(knearneigh(xy, longlat=TRUE))
map_nb <-make.sym.nb(map_nb)
map_nb

Warning message:
In knearneigh(xy, longlat = TRUE) : knearneigh: identical points found
Neighbour list object:
Number of regions: 3427 
Number of nonzero links: 4798 
Percentage nonzero weights: 0.04085376 
Average number of links: 1.400058

Create a spatial weights file

map_w <- nb2listw(map_nb,glist=NULL,style="W",zero.policy=FALSE)
map_w

Characteristics of weights list object:
Neighbour list object:
Number of regions: 3427 
Number of nonzero links: 4798 
Percentage nonzero weights: 0.04085376 
Average number of links: 1.400058 

Weights style: W 
Weights constants summary:
     n       nn   S0       S1       S2
W 3427 11744329 3427 5149.583 14944.42

Create a non row-standardized "listw" object - for the Getis ord G analysis

map_b <- nb2listw(map_nb, glist=NULL, style="B", zero.policy=FALSE)
map_b

Characteristics of weights list object:
Neighbour list object:
Number of regions: 3427 
Number of nonzero links: 4798 
Percentage nonzero weights: 0.04085376 
Average number of links: 1.400058 

Weights style: B 
Weights constants summary:
     n       nn   S0   S1    S2
B 3427 11744329 4798 9596 31432

Global spatial correlation:local geary's c

Rank boats by size/motor power
- Motor HP
```
gry.test(spatial.df$motor_hp1)      
```
  data.name Geary C statistic Expectation Variance p.value
  
  x
  
  weights: map_w
  
  0.226087563426783 1 0.000461813353282008 2.62384567961525e-284
- Bottom depth
```
gry.test(spatial.df$depth_m)      
```
  data.name Geary C statistic Expectation Variance p.value
  
  x
  
  weights: map_w
  
  0.184121438849991 1 0.000668255069430153 6.28123697497004e-219
- Distance from coast
```
gry.test(spatial.df$shoredist_km)      
```
  data.name Geary C statistic Expectation Variance p.value
  
  x
  
  weights: map_w
  
  0.00294931327704832 1 0.00082664039605247 8.34128907406172e-264
- Boat length
```
gry.test(spatial.df$boat_length1)      
```
  data.name Geary C statistic Expectation Variance p.value
  
  x
  
  weights: map_w
  
  0.187658371162682 1 0.000489304800176196 1.51669871424373e-295
- Total catch
```
gry.test(spatial.df$crew_size)      
```
  data.name Geary C statistic Expectation Variance p.value
  
  x
  
  weights: map_w
  
  0.344894015401117 1 0.000815596346756378 9.49338292882318e-117

Spatial patterns for mesh size distributions

Prepare the data

Create nearest neighbour and spatial weights for geary C testing spatial auto-correlation for mesh sizes

Subsetting the data to exclude the NA values for mesh sizes; can't run the geary's c test or getis ord because of the NAs. Create nearest neighbor file so that each feature is assessed within the spatial context of a fixed number of its closest neighbours.

spat.df <- subset(spatial.df, !is.na(mesh1mm))
xy<-cbind(spat.df$long, spat.df$lat)
colnames(xy) <- c("LONG","LAT")
plot(xy)
map_nb <-knn2nb(knearneigh(xy, longlat=TRUE))
map_nb <-make.sym.nb(map_nb)
map_w <- nb2listw(map_nb, glist=NULL, style="W", zero.policy=FALSE)
map_b <- nb2listw(map_nb, glist=NULL, style="B", zero.policy=FALSE)

Warning message:
In knearneigh(xy, longlat = TRUE) : knearneigh: identical points found

Run the Geary test

geary.test(spat.df$mesh1mm, map_w) # removed na.action=na.omit)
geary.test(spat.df$mesh1mm, map_w, zero.policy=NULL, 
alternative = "greater", spChk=NULL) # na.action=na.exclude, , p.adjust.method="bh"

   Geary C test under randomisation

data:  spat.df$mesh1mm 
weights: map_w 

Geary C statistic standard deviate = 23.396, p-value < 2.2e-16
alternative hypothesis: Expectation greater than statistic
sample estimates:
Geary C statistic       Expectation          Variance 
     0.4553284724      1.0000000000      0.0005420015

    Geary C test under randomisation

data:  spat.df$mesh1mm 
weights: map_w 

Geary C statistic standard deviate = 23.396, p-value < 2.2e-16
alternative hypothesis: Expectation greater than statistic
sample estimates:
Geary C statistic       Expectation          Variance 
     0.4553284724      1.0000000000      0.0005420015

Spatial autocorrelation test on categorical data:join counts

The joincount.multi function is used to tally join counts between same-colour and different colour spatial objects

For craft type

spat.df$boat_type1.f <- as.factor(spat.df$boat_type1)
jcmulti.craft <- joincount.multi(spat.df$boat_type1.f, map_w, zero.policy = FALSE, spChk = NULL, adjust.n=TRUE)
jcmulti.craft <- as.data.frame(jcmulti.craft)
colnames(jcmulti.craft)[4] <- "z.value"
jcmulti.craft$p.value <- pnorm(-abs(jcmulti.craft$z.value))
jcmulti.craft$adjusted.p.value <- p.adjust(jcmulti.craft$p.value, "fdr")
write.csv(jcmulti.craft, file="./Results/jc_multi_craft.csv")
jcmulti.craft

	Joincount	Expected	Variance	z.value	p.value	adjusted.p.value
frp:frp	448.333333333333	221.529652351738	77.4143727087576	25.7774111903593	7.94552119506807e-147	2.54256678242178e-146
frp_big:frp_big	4.16666666666667	0.184049079754601	0.135801478161905	10.8072788834754	1.58981043503155e-27	3.1796208700631e-27
kattumaram:kattumaram	267.333333333333	82.4937189599766	40.3033663344659	29.1155093937937	1.14224765298328e-186	4.5689906119331e-186
trawl:trawl	509.916666666667	217.233420975752	76.6089842455549	33.4393641263136	1.83720717837875e-245	1.469765742703e-244
vallam:vallam	64.4166666666667	4.91849255039439	3.35550846525004	32.4806318382891	1.00090644424911e-231	5.33816770266191e-231
frp_big:frp	23.0416666666667	12.9570552147239	6.61538223381007	3.92086327961169	4.41161571229531e-05	5.42968087667115e-05
kattumaram:frp	179.916666666667	270.658486707566	124.031469638634	-8.14782215570812	1.85268372545502e-16	2.96429396072803e-16
kattumaram:frp_big	1.5	7.90885188431201	4.75839387704019	-2.93798815597406	0.00165174821937064	0.00188771225070931
trawl:frp	122.083333333333	439.100204498978	186.209522591467	-23.2317235995092	1.08853609161107e-119	2.90276291096287e-119
trawl:frp_big	5.41666666666667	12.8308501314636	6.58016639394049	-2.89031309527329	0.00192429154716521	0.00205257765030955
trawl:kattumaram	33.5416666666667	268.022202746129	123.208989408158	-21.1244592533643	2.37010962524273e-99	5.41739342912623e-99
vallam:frp	10.9166666666667	66.2249488752556	33.0959668276582	-9.61396715004607	3.49031960107625e-22	6.20501262413556e-22
vallam:frp_big	0.75	1.93514460999124	1.37288140379489	-1.01147425810669	0.155894747798648	0.155894747798648
vallam:kattumaram	6.25	40.4230207420391	23.603387262254	-7.03390023473145	1.00419376497186e-12	1.46064547632271e-12
vallam:trawl	34.4166666666667	65.5799006719252	32.911581659916	-5.43210484651395	2.7846590998081e-08	3.71287879974413e-08
Jtot	417.833333333333	1185.64066608238	255.932687124977	-47.994268540164	0	0

For type of net

spat.df$fishery.f <- as.factor(spat.df$fishery)
jcmulti.fy <- joincount.multi(spat.df$fishery.f, map_w, zero.policy = FALSE, spChk = NULL, adjust.n=TRUE)
jcmulti.fy <- as.data.frame(jcmulti.fy)
colnames(jcmulti.fy)[4] <- "z.value"
jcmulti.fy$p.value <- pnorm(-abs(jcmulti.fy$z.value))
jcmulti.fy$adjusted.p.value <- p.adjust(jcmulti.fy$p.value, "fdr")
write.csv(jcmulti.fy, file="./Results/jc_multi_fy.csv")
jcmulti.fy

	Joincount	Expected	Variance	z.value	p.value	adjusted.p.value
Gillnet:Gillnet	750.416666666667	451.184049079755	100.539669661172	29.8428437238321	5.43584048372764e-196	3.98628302140027e-195
Lift:Lift	0	0.00175284837861525	0.00131492421018604	-0.0483385913605902	0.480723199514433	0.480723199514433
Line:Line	18.5833333333333	0.685363716038563	0.497070405936588	25.3860313969905	1.79877308724197e-142	6.59550131988724e-142
Ringseine:Ringseine	64.4166666666667	6.59596844872918	4.42967172557744	27.4724606738747	1.87317431447735e-166	8.24196698370034e-166
Scoopnet:Scoopnet	32.75	2.64241893076249	1.85076302389712	22.1309704005577	7.95580419290557e-109	2.50039560348461e-108
Trawl:Trawl	491.166666666667	226.231375985977	78.2758256129964	29.9451148981374	2.54694182947884e-197	2.80163601242673e-196
Lift:Gillnet	4	2.05433829973707	0.845227440513777	2.11631610056557	0.0171589661495446	0.0290382504069216
Line:Gillnet	10.9166666666667	35.4373356704645	14.5301549943358	-6.43275767288728	6.26546737717324e-11	1.5315586921979e-10
Line:Lift	0	0.0806310254163015	0.0594444400404008	-0.330709417787497	0.370431993931938	0.407475193325132
Ringseine:Gillnet	30.6666666666667	109.393514460999	44.5118913994447	-11.8000769218461	1.94977417202781e-32	5.36187897307646e-32
Ringseine:Lift	0	0.248904469763365	0.176588500705512	-0.592313301977417	0.276820405031792	0.338336050594413
Ringseine:Line	2	4.29360210341805	2.99800946946988	-1.32465132456023	0.0926434122381213	0.119891474661098
Scoopnet:Gillnet	47.0416666666667	69.3339176161262	28.3291965521648	-4.18829047908432	1.4053181308645e-05	2.57641657325157e-05
Scoopnet:Lift	0	0.157756354075372	0.114295899612428	-0.466628795609195	0.320382740657809	0.370969489182726
Scoopnet:Line	0.416666666666667	2.72129710780018	1.93974509733649	-1.65473694191114	0.0489889350189446	0.0718504380277855
Scoopnet:Ringseine	4.58333333333333	8.40052585451359	5.77003040811361	-1.58911385194109	0.0560173455123581	0.0770238500794924
Trawl:Gillnet	172.291666666667	639.412795793164	245.846135938948	-29.7918737356214	2.48910962054384e-195	1.36901029129911e-194
Trawl:Lift	0	1.45486415425066	0.742551175085095	-1.68833743001056	0.0456732370014522	0.0717722295737106
Trawl:Line	24.75	25.0964066608238	12.6919130236038	-0.0972350161244822	0.461269880469087	0.480723199514433
Trawl:Ringseine	40.5833333333333	77.4715162138475	38.3785965983598	-5.95446459712357	1.30462300739861e-09	2.60924601479723e-09
Trawl:Scoopnet	17.4166666666667	49.1016652059597	24.5993807335103	-6.38839269347037	8.38191977461669e-11	1.84402235041567e-10
Jtot	354.666666666667	1024.65907099036	258.30117574792	-41.6875802832853	0	0

Getis-Ord G-test

The Getis-Ord G statistic differs from Moran's I or Geary C by the fact that it makes a difference between high-high correlation and low-low correlation (both of which Moran's I treats as positive autocorrelation) only binary non row-standardized weight matrices work for the G-test

Global G test

globalG.test(spat.df$mesh1mm, map_b)

   Getis-Ord global G statistic

data:  spat.df$mesh1mm 
weights: map_b 

standard deviate = 17.778, p-value < 2.2e-16
alternative hypothesis: greater
sample estimates:
Global G statistic        Expectation           Variance 
      4.893520e-04       4.092027e-04       2.032479e-11

Local G test

An example to plot the localG is here: http://isites.harvard.edu/fs/docs/icb.topic923307.files/R code for Lab Ex 6.txt

For mesh size

Gmesh <- localG(spat.df$mesh1mm, map_b, zero.policy=NULL, spChk=NULL)
summary(Gmesh)

     Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 
-2.029673 -0.720682 -0.404282  0.004912  0.756180  5.822354

For crew size

Gcrew <- localG(spat.df$crew_size, map_b)
summary(Gcrew)

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
-1.02179 -0.47916 -0.25810 -0.01024 -0.03710  8.06191

plotting distance violations

Mechanised craft

mech <- subset(spatial.df, sector == "mech")
mechboats<-c('trawl' = "Trawler", 'vallam' = "Vallam")
sp <- ggplot(mech, aes(x=shoredist_km,y=depth_m, colour = factor (distvio))) + geom_point(shape=21, position=position_jitter(width=0.5,height=0.5))
sp
a<-sp + scale_colour_manual(name="Violation of shore distance (km)", labels=c("> 5.556km","< 5.556km"), values = c("dark grey", "black"))
b<-a + facet_grid(region ~ boat_type1, labeller = labeller(boat_type1 = as_labeller(mechboats)))
b
c<-b + ggtitle("Violation of fishing 3 nautical miles (5.556 km) from the shore by mechanized craft") + xlab("Distance from shore (km)") + ylab("Depth of substrate(m)")
d<-c + theme(strip.text.x = element_text(size=12, face="bold"), strip.text.y = element_text(size=12, face="bold"))
e<-d + theme(legend.position = "bottom")
f<-e + theme(axis.text = element_text(size=12, face="bold"))
g<-f + theme(axis.title.x = element_text(face="bold", size=14), axis.title.y = element_text(face="bold", size=14))
g

Mesh size violations

mesh <- ggplot(spatial.df, aes(x=shoredist_km, y=depth_m, colour = factor (meshvio))) + geom_point(shape=21, position=position_jitter(width=0.5,height=0.5)) + facet_wrap(~region)
c<-mesh + ggtitle("Violation of gear mesh sizes") + xlab("Distance from shore (km)") + ylab("Depth of substrate(m)")
d<-c + theme(strip.text.x = element_text(size=12, face="bold"), strip.text.y = element_text(size=12, face="bold"))
e<-d + theme(legend.position = "bottom")
f<-e + theme(axis.text = element_text(size=12, face="bold"))
g<-f + theme(axis.title.x = element_text(face="bold", size=14), axis.title.y = element_text(face="bold", size=14))
g + scale_colour_manual(name="Violation of gear mesh sizes(mm)", labels=c("No","Yes"), values = c("grey", "black")) + scale_size_continuous(name  ="Mesh sizes(mm)")
## creating groups to bin mesh sizes going from 0  to 120 in increments of 20.
spatial.df$mesh1mmcat<-cut(spatial.df$mesh1mm, seq(0,120,20))
## renaming the facet panels
meshcat<-c(
  '(0,20]' = "0-20",
  '(20,40]' = "20-40",
  '(40,60]' = "40-60",
  '(60,80]'= "60-80",
  '(80,100]'="80-100",
  '(100,120]'="100-120"
)
## plotting distribution of mesh sizes
p<-ggplot(spatial.df,aes(shoredist_km, depth_m, colour = factor(fishery))) +
    geom_point() +
    facet_wrap(~mesh1mmcat, scales = "free_x", labeller = labeller(mesh1mmcat = as_labeller(meshcat))) ## mesh1mmcat doesn't exit
q<- p + theme_gray() + geom_point(shape=21, position=position_jitter(width=0.5,height=0.5))
r<- q + ggtitle("Cumulative distribution of gear mesh sizes") + xlab("Distance from shore(km)") + ylab("Depth of substrate(m)")
s<- r + theme(legend.position = "bottom") + scale_colour_discrete(name  ="Fishery", 
                                                                  breaks=c("Gillnet", "Lift","Line", "Ringseine", "Scoopnet","Trawl"),
                                                                  labels=c("Gill net", "Lift net","Line", "Ring seine", "Scoop net","Trawl"))
t<-s + theme(axis.title.x = element_text(face="bold", size=14), axis.title.y = element_text(face="bold", size=14))
u<-t + theme(axis.text = element_text(size=12))
v<-u + theme(plot.title = element_text(size = 16, face = "bold")) + theme(strip.text.x = element_text(size=12, face="bold"))
x<-v + theme(legend.text = element_text(size=12)) + theme(legend.title = element_text(size=12, face = "bold"))
x

Generate plots

generate histogram of distance from shoreline

distance of all craft

adapted from http://www.fromthebottomoftheheap.net/2013/09/09/preparing-figures-for-plos-one-with-r/

ggplt <- ggplot(data=fdf, aes(fdf$shoredist_)) +
  geom_vline(xintercept = 5.556, colour="black", linetype = "longdash") +
  geom_histogram(breaks=seq(1, 45, by =0.5), 
                 ## col="gray5", # uncomment if you want grayscale
                 aes(fill=..count..)) +
  labs(x="Distance (km)", y="Count") +
  ##    theme(plot.margin = unit(c(5,4,0.05,0.05), "in")) +
  scale_fill_gradient("Count", low = "gray60", high = "black")
  ggplt

distance per craft type

ggplot(data=fdf, aes(fdf$shoredist_)) +
  facet_wrap (~boat_type1)+
  geom_vline(xintercept = 5.556, colour="black", linetype = "longdash") + 
  geom_histogram(breaks=seq(1, 45, by=1),
                 ## col="gray5",  # uncomment if you want grayscale
                 aes(fill=..count..)) +
  labs(x="Distance (km)", y="Count") +
  scale_fill_gradient("Count", low = "gray60", high = "black")

Depth per craft type

ggplot(data=fdf, aes(fdf$depth_m)) +
  facet_wrap (~boat_type1)+
  geom_histogram(breaks=seq(1, 115, by=2),
                 ## col="gray5",  # uncomment if you want grayscale
                 aes(fill=..count..)) +
  labs(x="Depth (m)", y="Count") +
  scale_fill_gradient("Count", low = "gray60", high = "black")

Processing graphs and data on violations

Subsetting the data by district, type of craft and class of craft

District

pon<-subset (fdf, region == "Pondicherry")
cud<- subset (fdf,region == "Cuddalore")
vil<- subset (fdf, region == "Villupuram")
tn<-rbind(cud, vil)

Subsetting craft type: mechanised v/s artisanal for Pondy

trawl<-subset (pon, boat_type1 == "Trawl")
frp<-subset(pon,boat_type1 == "FRP")
vallam<-subset(pon,boat_type1 == "Vallam")
kat<-subset(pon,boat_type1 == "Kattumaram")
frpb<-subset(pon,boat_type1 == "Big FRP")
ponart<-rbind(frp, kat, frpb)
ponmech<-rbind(vallam,trawl)

Subsetting craft type: mechanised v/s artisanal for tn

tntrawl <- subset(tn, boat_type1 == "Trawl")
tnfrp <- subset(tn,boat_type1 == "FRP")
tnvallam <- subset(tn,boat_type1 == "Vallam")
tnkat <- subset(tn,boat_type1 == "Kattumaram")
tnfrpb <- subset(tn,boat_type1 == "Big FRP")
tnart <- rbind(tnfrp, tnkat, tnfrpb)
tnmech <- rbind(tnvallam,tntrawl)

Plotting for distance violations - PONDICHERRY

plt <- ggplot(ponmech,aes(x=shoredist_,y=depth_m, size=crew_size, colour = factor(shoredist_>5.556))) +
 theme_gray() +
 geom_point(shape=21, position=position_jitter(width=0.5,height=0.5)) + 
 facet_wrap (~boat_type1) +
 ggtitle("MFRA Violoations of Distance of Fishing from Shore: Pondicherry") +
 xlab("Distance from shore (km)") +
 ylab("Depth of substrate(km)") +
 theme(legend.title = element_text(size = 12), plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
 theme(legend.position="right")

plt1.bw <- plt + scale_colour_manual(name="Violations",
  labels=c("< 5.556km","> 5.556km"), values = c("black", "darkgrey")) +
  theme(legend.title = element_text(size = 12), plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
  scale_size_continuous(name  ="Size of crew") 
plt1.bw

plt1.col <- plt + scale_colour_manual(name="Violations",
                                      labels=c("< 5.556km","> 5.556km"), values = c("red", "darkgreen")) +
    theme(legend.title = element_text(size = 12)) +
    scale_size_continuous(name  ="Size of crew")
plt1.col

Plotting for distance violations - TAMILNADU

plt <- ggplot(tnmech,aes(x=shoredist_,y=depth_m, size=crew_size, colour = factor(shoredist_>5.556)))+
geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) +
facet_wrap (~boat_type1) +
ggtitle("MFRA Violoations of Distance of Fishing from Shore: Tamil Nadu") +
xlab("Distance from shore (Km)") +
ylab("Depth of substrate(m)") +
theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5))

plt2.col <- plt  + scale_colour_manual(name="Violations",
labels=c("< 5.556km","> 5.556km"), values = c("red", "darkgreen")) +
scale_size_continuous(name  ="Size of crew")
plt2.col

plt2.bw <- plt  + scale_colour_manual(name="Violations",
labels=c("< 5.556km","> 5.556km"), values = c("black", "darkgrey"))+
scale_size_continuous(name  ="Size of crew")
plt2.bw

mylegend <- g_legend(plt1.col)
plot.cmd <- grid.arrange(arrangeGrob(plt1.col + theme(legend.position="none")+ xlab(""),
                      plt2.col + theme(legend.position="none"), nrow=2),
          mylegend, ncol=2, widths=c(40, 6))
plot.cmd

 mylegend <- g_legend(plt1.bw)
 plot.cmd <- grid.arrange(arrangeGrob(plt1.bw + theme(legend.position="none")+ xlab(""),
                       plt2.bw + theme(legend.position="none"), nrow=2),
           mylegend, ncol=2, widths=c(40, 6))
flnm <- "./Results/violations_distance_tn_pdy_bw"
eps.tiff(plot.cmd, flnm, width.in = 9, height.in = 7.5)
plot.cmd

plotting for mesh violations - TAMILNADU (general 10 mm limitation)

tnmesh1<-subset(tn, net_type != "line")
plt.tn <- ggplot(tnmesh1,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>10))) +
 geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) +
 facet_wrap (~boat_type1, nrow=1)+ggtitle("MFRA Tamil Nadu: Violations of Gear Mesh Sizes") +
 xlab("Distance from shore (Km)") +
 ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
 plt.tn.col <- plt.tn +  scale_colour_manual(name="Violations",
                                   labels=c("< 10mm","> 10mm"), values = c("red", "darkgreen")) +
 scale_size_continuous(name  ="Mesh size(mm)")
 plt.tn.col

plt.tn.bw <- plt.tn.col +  scale_colour_manual(name="Violations",
                                  labels=c("< 10mm","> 10mm"), values = c("black", "darkgrey")) +
 scale_size_continuous(name  ="Mesh size(mm)")
 plt.tn.bw

plotting for mesh violations - PONDICHERRY (general 25 mm limitation)

   ponmesh1 <- subset(pon, net_type != "line")
   plt.pon <- ggplot(ponmesh1,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>25))) +
   geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) +
    facet_wrap (~boat_type1, nrow=1)+ggtitle("MFRA Pondicherry: Violoations of Gear Mesh Sizes") +
    xlab("Distance from shore (Km)") +
    ylab("Depth of substrate(m)") +
    theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
plt.col.pon <- plt.pon +  scale_colour_manual(name="Violations",
                                      labels=c("< 25mm","> 25mm"), values = c("red", "darkgreen")) +
    scale_size_continuous(name  ="Mesh size(mm)")
plt.col.pon

plt.bw.pon <- plt.pon +  scale_colour_manual(name="Violations",
                                     labels=c("< 25mm","> 25mm"), values = c("black", "darkgrey")) +
    scale_size_continuous(name  ="Mesh size(mm)")
plt.bw.pon

plot.cmd <- grid.arrange(arrangeGrob(plt.bw.pon + theme(legend.position="right")+ xlab("") ,
                                     plt.tn.bw + theme(legend.position="right"), nrow=2))
flmn <- "./Results/all_violations_mesh_tn_pdy_col"
eps.tiff(plot.cmd, flnm, width.in = 9, height.in = 7.5)
plot.cmd

Subsetting the data: gillnet mesh violations - PONDICHERRY

pgill1 <- subset (pon, net_type == "multigill")
pgill2 <- subset (pon,net_type == "monogill")
pongill <- rbind(pgill1, pgill2)
tngill1 <- subset (tn, net_type == "multigill")
tngill2 <- subset (tn,net_type == "monogill")
tngill <- rbind(tngill1, tngill2)
allgill <- rbind(pongill, tngill)

plotting for gillnet violations

plt <- ggplot(pongill,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>25))) +
 geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) + facet_wrap (~boat_type1) +
 ggtitle("Gill net mesh sizes used for fishing in Pondicherry") +
 xlab("Distance from shore (km)") +
 ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) 
 plt.col <- plt + scale_colour_manual(name="Violations",
                                  labels=c("< 25mm","> 25mm"), values = c("red", "darkgreen")) +
 scale_size_continuous(name  ="Mesh size(mm)")
plt.col # colour plot

plt.bw <- plt + scale_colour_manual(name="Violations",
 labels=c("< 25mm","> 25mm"), values = c("darkgrey", "black")) +
 scale_size_continuous(name  ="Mesh size(mm)")
plt.bw # grayscale plot

Plotting for violations across districts for pondy and TN MFRA

plt <- ggplot(allgill,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>25))) +
 geom_point(aes(shape=factor(region)), position=position_jitter(width=1,height=.5))+
 labs(shape="District")+
 facet_wrap (~boat_type1) +
 ggtitle("Gill net mesh size violations: Pondicherry MFRA") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
pltpdy.col <- plt + scale_colour_manual(name="Violations",
                                     labels=c("< 25mm","> 25mm"), values = c("red", "darkgreen")) +
                                     theme(legend.title = element_text(size = 12)) +
                                     scale_size_continuous(name  ="Mesh size (mm)") +
                                     scale_shape_discrete(solid=F)
 pltpdy.col # colour plot

pltpdy.bw <- plt + scale_colour_manual(name="Violations",
                                    labels=c("< 25mm","> 25mm"), values = c("black", "darkgrey")) +
 theme(legend.title = element_text(size = 12)) +
 scale_size_continuous(name  ="Mesh size (mm)")+
 scale_shape_discrete(solid=F)
 pltpdy.bw # grayscale plot

plt <- ggplot(allgill,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>10))) +
 geom_point(aes(shape=factor(region)), position=position_jitter(width=0.5,height=0.5))+
 labs(shape="District")+
 facet_wrap (~boat_type1) +
 ggtitle("Gill net mesh size violations: Tamil Nadu MFRA") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5))
 ## guides(fill=guide_legend(nrow=3,byrow=TRUE))
 plttn.col <- plt + scale_colour_manual(name="Violations",
                                    labels=c("< 10mm","> 10mm"), values = c("red", "darkgreen")) +
 scale_size_continuous(name  ="Mesh size (mm)") + 
 scale_shape_discrete(solid=F)
 plttn.col # colour plot

plttn.bw <- plt + scale_colour_manual(name="Violations",
                                   labels=c("< 10mm","> 10mm"), values = c("black", "darkgrey")) +
 scale_size_continuous(name  ="Mesh size (mm)") +
 scale_shape_discrete(solid=F) 
 plttn.bw # grayscale plot

Merging the two plots

mylegend <- g_legend(pltpdy.col)
pltpdy.col <- pltpdy.col + xlab("")
grid.arrange(arrangeGrob(pltpdy.col + theme(legend.position="none")+ xlab("") ,
    plttn.col + theme(legend.position="none"), nrow=2),
    mylegend, ncol=2, widths=c(40, 6))

mylegend <- g_legend(pltpdy.bw) 
grid.arrange(arrangeGrob(pltpdy.bw + theme(legend.position="none")+ xlab(""),
  plttn.bw + theme(legend.position="none"), nrow=2),
  mylegend,  ncol=2, widths=c(40, 6))

Plotting trawl net mesh violations - PONDICHERRY

mtrawl<-subset(ponmech, net_type == "multitrawl")
mtrawler<-subset(ponmech, boat_type1 == "Trawl")
ca <- ggplot(mtrawl,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(mesh1mm>37))) +
 geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) +
 facet_wrap (~boat_type1)+ggtitle("Pondicherry MFRA Violations of Trawl Net Mesh Sizes") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
 scale_colour_discrete(name="Violations",labels=c("< 37mm","> 37mm")) +
 scale_size_continuous(name ="Mesh size(mm)")
ca

plotting for multiday fishing - PONDICHERRY

 da <- ggplot(mtrawler,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(no_days > 1))) +
  geom_point(shape=1, position=position_jitter(width=0.5,height=0.5)) +
  ggtitle("Pondicherry MFRA Violoations Multiday Trawl Fishing") +
  xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
  theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
  scale_colour_discrete(name="MFRA violation of multiday fishing",labels=c("single day fishing","multiday fishing")) +
  scale_size_continuous(name  ="Mesh sizes used")
da # plot da

Plotting for multiday fishing - TAMILNADU

ea <- ggplot(tnmech,aes(x=shoredist_,y=depth_m, size=mesh1mm, colour = factor(no_days > 1))) +
 geom_point(shape=1,  position=position_jitter(width=0.5,height=0.5)) +
 facet_wrap (~boat_type1)+ggtitle("Tamil Nadu MFRA Violoations Multiday Trawl Fishing") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
 scale_colour_discrete(name="Violations",labels=c("Single day fishing","multiday fishing")) +
 scale_size_continuous(name  ="Mesh sizes used")
 ea

Plotting for motor HP violations - TAMILNADU

ij <- ggplot(tnmech,aes(x=shoredist_,y=depth_m, size=motor_hp1,colour = factor(motor_hp1>150))) +
geom_point(shape=1,  position=position_jitter(width=1, height=1.5)) +
 facet_wrap (~boat_type1)+ggtitle("Tamil Nadu MFRA Violation of Engine Horsepower in Mechanised Craft") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
 scale_colour_discrete(name="Violations", labels=c("Mechanised <150 HP","Deep sea >150 HP"))+
 scale_size_continuous(name ="Motor HP range")
ij

Plotting for motor HP - PONDICHERRY

j <- ggplot(ponmech,aes(x=shoredist_,y=depth_m, size=motor_hp1, colour = factor(motor_hp1>150))) +
 geom_point(shape=1, position = position_jitter(width=1, height=1.5)) +
 facet_wrap (~boat_type1) +
 ggtitle("Pondicherry MFRA Violation of Engine Horsepower in Mechanised Craft") +
 xlab("Distance from shore (km)") + ylab("Depth of substrate(m)") +
 theme(legend.position="right", plot.title = element_text(size = 12, face = "plain", hjust = 0.5)) +
 scale_colour_discrete(name="Violations", labels=c("Mechanised <150 HP","Deep sea >150 HP")) +
 scale_size_continuous(name  ="Motor HP range")
j

Mapping

Get the basemaps and panels

Stamen map

You need to be on-line for this map to work as it fetches the map from the OSM server.

pdymap.bw <- get_map(location = c(lon = 79.86308, lat=12.0), zoom = 9, source = "stamen", maptype = "toner-lite") 
plot(pdymap.bw)

Plot of distances

fdf$zone[fdf$shoredist_>5.556] <- "Non-Artisanal > 3nm (5.556km)" # changed from 3km to 3nm or 5.556km
fdf$zone[fdf$shoredist_<=5.556] <- "Artisanal <= 3nm (5.556km)"
dist.map <- qmplot(long, lat, data = fdf, maptype = "toner-lite", colour =zone, extent = "panel") +
scale_colour_manual(values = c("black", "darkgrey")) +
theme(plot.caption = element_text(colour="black", size=6, face="bold")) +
labs(x = "Longitude", y = "Latitude", colour = "Zone", caption = "NOT TO SCALE")+
scalebar(fdf,dist = 10, dd2km = TRUE, model = 'WGS84', height = 0.0075, st.size = 2.5)
plot(dist.map)

Plot panels of different variables

We use lapply to run this function quickly using the script below.

panels <- list("net_type", "crew_size", "boat_type1", "boat_lengt", "motor_hp1", "boat_lengt", "substrate", "mesh1mm", "catchwt_kg", "target_cat", "survey_mon", "no_days", "CraftClass","zone")
lapply(X = panels, FUN = qmplot.panels)

Here is an example of the output for a single map:

qmplot.panels("net_type")

GIS data processing using GRASS and R

We use the rgrass7 library to process the GIS data. Most of the commands run functions under a version of apply to loop them over lists. We've provided some output examples where practical.

Create a 10km grid over the data and clean up legacy files

execGRASS("g.mapset", mapset="craftdist") # ensure you're in the correct mapset
execGRASS("g.region", vector="BtDistUTM@PERMANENT", res="1000") # select computational extent
execGRASS("v.mkgrid", parameters = list(map="onekmgrid", box=c(1000,1000)), flags = "overwrite") # create 1km grid extending over region
execGRASS("g.remove", type="vector", name="frp,frp_big,kattumaram,trawl,vallam,frp_big_boat_type1,frp_boat_type1,kattumaram_boat_type1,trawl_boat_type1,vallam_boat_type1", flags = "f") # clean up

Create subsets of maps

This calls the function subset.map which creates new layers based on input list by subsetting maps through v.extract

run for fishery

fishery.maps <- list("Gillnet", "Lift", "Line", "Ringseine", "Scoopnet", "Trawl") # list of maps to be created
lapply(X = fishery.maps, FUN = subset.map, colhd = "fishery") # run the function

run for boat type

craft.maps <- as.list(as.character(unique(fdf$boat_type1)))
lapply(X = craft.maps, FUN = subset.map, colhd = "boat_type1")

run for class of craft

craftclass.maps <- as.list(as.character(unique(fdf$CraftClass)))
lapply(X = craftclass.maps, FUN = subset.map, colhd = "CraftClass")

Subset map based on mesh size

This may throw an error Column <meshsize> is already in the table. Skipping. You can ignore this.

execGRASS("g.mapset", mapset="PERMANENT") #switch to PERMANENT mapset
try(execGRASS("v.db.addcolumn", map="BtDistUTM@PERMANENT", columns="meshsize INT"))#add integer column
execGRASS("v.db.update", map="BtDistUTM@PERMANENT", column="meshsize", query_column="mesh1mm") #copy values
execGRASS("g.mapset", mapset="craftdist") #switch back to craftdist mapset
lwr.mesh <- as.list(seq(0, 110, by= 20)) 
upr.mesh <- as.list(seq(20, 120, by= 20))
nm.mesh <- paste0("mesh",seq(20, 120, by=20))
execGRASS("g.remove", type = "vector", pattern = "mesh*", flags = "f", intern=TRUE) # remove existing maps HERE
mapply(FUN = subsetmesh, lwr=lwr.mesh, upr=upr.mesh, nm=nm.mesh)

Process craft classes

CraftClass.maps <- as.list(execGRASS("g.list", type = "vector", pattern = "*CraftClass", mapset = "craftdist", intern = T))
CraftClass.facets <- gsub("_CraftClass", "", CraftClass.maps)
CraftClass.df <- do.call("rbind", mapply(read.vect.list, CraftClass.maps, CraftClass.facets, SIMPLIFY = FALSE))
qmplot.dens2d.mesh(CraftClass.df, "./Results/CraftClassMeshSizes.tiff")
qmplot.dens2d.mesh(CraftClass.df, "./Results/CraftClassMeshSizes.eps")
qmplot.dens2d.mesh(CraftClass.df, "./Results/CraftClassMeshSizes.png")

Build quadrats around fisheries

net.maps <- as.list(execGRASS("g.list", type = "vector", pattern = "*fishery", mapset = "craftdist", intern = T))
net.map.facets <- as.list(gsub("_fishery","", net.maps))
net.df <- do.call("rbind", mapply(read.vect.list, net.maps, net.map.facets, SIMPLIFY = FALSE))
qmplot.dens2d.mesh(net.df, "./Results/NetTypesMeshSizes.tiff")
qmplot.dens2d.mesh(net.df, "./Results/NetTypesMeshSizes.eps")
qmplot.dens2d.mesh(net.df, "./Results/NetTypesMeshSizes.png")

Build quadrats around mesh sizes

mesh.maps <- as.list(execGRASS("g.list", type = "vector", pattern = "mesh*", intern = T))
mesh.maps <- mesh.maps[c(3,4,5,6,1,2)]
mesh.map.facets <- as.list(paste(seq(0, 110, 20), "to", seq(20, 120, 20)))
mesh.df <- do.call("rbind", mapply(read.vect.list, mesh.maps, mesh.map.facets, SIMPLIFY = FALSE))
mesh.df$facet <- factor(mesh.df$facet, levels = c("0 to 20","20 to 40","40 to 60","60 to 80","80 to 100","100 to 120")) # fix order of facets
qmplot.dens2d.mesh(mesh.df, "MeshSizeClassesMeshSizes.tiff")
qmplot.dens2d.mesh(mesh.df, "MeshSizeClassesMeshSizes.eps")
qmplot.dens2d.mesh(mesh.df, "MeshSizeClassesMeshSizes.png")

boat.maps <- as.list(execGRASS("g.list", type = "vector", pattern = "*boat_type1", mapset = "craftdist", intern = T))
boat.map.facets <- as.list(gsub("_boat_type1","", boat.maps))
boats.df <- do.call("rbind", mapply(read.vect.list, boat.maps, boat.map.facets, SIMPLIFY = FALSE))
qmplot.dens2d.mesh(boats.df, "BoatTypeMeshSizes.tiff")
qmplot.dens2d.mesh(boats.df, "BoatTypeMeshSizes.eps")
qmplot.dens2d.mesh(boats.df, "BoatTypeMeshSizes.png")

Generate interactive maps

Craft and mesh size

p <- qmplot(x = long, y = lat, data = boats.df, maptype = "toner-lite", extent = "panel", zoom=10, geom="text", label="") +
        geom_point(data = boats.df, aes(x = long, y = lat, colour=meshsize, shape=boat_type1)) +
        stat_density2d(data = boats.df, aes(x = long, y = lat)) +
        labs(
            x = NULL,
            y = NULL ,
            title = "Craft and Mesh Sizes",
            colour = "Mesh \nSize (mm)",
            shape = "Craft Type",
            caption = NULL 
        ) +
    coord_equal()
p <- ggplotly(p)
htmlwidgets::saveWidget(p, "CraftMeshsizePlotly.html")

Coordinate system already present. Adding new coordinate system, which will replace the existing one.

Mesh classes and their distribution

p <- qmplot(x = long, y = lat, data = mesh.df, maptype = "toner-lite", extent = "panel", zoom=10, geom="text", label="") +
        geom_point(data = mesh.df, aes(x = long, y = lat, colour=meshsize, shape=facet)) +
        stat_density2d(data = mesh.df, aes(x = long, y = lat)) +
        labs(
            x = NULL,
            y = NULL ,
            title = "Mesh Size Class and Mesh Sizes",
            colour = "Mesh \nSize (mm)",
            shape = "Mesh Size \nClass",
            caption = NULL 
        ) +
    coord_equal()
p <- ggplotly(p)
htmlwidgets::saveWidget(p, "MeshSizeClassMeshsizePlotly.html")

Coordinate system already present. Adding new coordinate system, which will replace the existing one.

Net types and mesh sizes

p <- qmplot(x = long, y = lat, data = net.df, maptype = "toner-lite", extent = "panel", zoom=10, geom="text", label="") +
        geom_point(data = net.df, aes(x = long, y = lat, colour=meshsize, shape=facet)) +
        stat_density2d(data = net.df, aes(x = long, y = lat)) +
        labs(
            x = NULL,
            y = NULL ,
            title = "Net Types and Mesh Sizes",
            colour = "Mesh \nSize (mm)",
            shape = "Net Type",
            caption = NULL 
        ) +
    coord_equal()
p <- ggplotly(p)
htmlwidgets::saveWidget(p, "NetTypeMeshsizePlotly.html")

Coordinate system already present. Adding new coordinate system, which will replace the existing one.

Craft classes and mesh sizes

p <- qmplot(x = long, y = lat, data = CraftClass.df, maptype = "toner-lite", extent = "panel", zoom=10, geom="text", label="") +
        geom_point(data = CraftClass.df, aes(x = long, y = lat, colour=meshsize, shape=facet)) +
        stat_density2d(data = CraftClass.df, aes(x = long, y = lat)) +
        labs(
            x = NULL,
            y = NULL ,
            title = "Class of Craft and Mesh Sizes",
            colour = "Mesh \nSize (mm)",
            shape = "Craft Class",
            caption = NULL 
        ) +
    coord_equal()
p <- ggplotly(p)
htmlwidgets::saveWidget(p, "CraftClassMeshsizePlotly.html")

Coordinate system already present. Adding new coordinate system, which will replace the existing one.

data.name	Geary C statistic	Expectation	Variance	p.value
x
weights: map_w
	0.226087563426783	1	0.000461813353282008	2.62384567961525e-284

data.name	Geary C statistic	Expectation	Variance	p.value
x
weights: map_w
	0.184121438849991	1	0.000668255069430153	6.28123697497004e-219

data.name	Geary C statistic	Expectation	Variance	p.value
x
weights: map_w
	0.00294931327704832	1	0.00082664039605247	8.34128907406172e-264

data.name	Geary C statistic	Expectation	Variance	p.value
x
weights: map_w
	0.187658371162682	1	0.000489304800176196	1.51669871424373e-295

data.name	Geary C statistic	Expectation	Variance	p.value
x
weights: map_w
	0.344894015401117	1	0.000815596346756378	9.49338292882318e-117

Code documentation for paper "Spatially Explicit Action Research for Coastal Fisheries Management"

Tara Lawrence and R.S. Bhalla

Table of Contents

Summary

Initialise

Set up the environment

Load the libraries

Datasets

Functions and initialisation scripts

Run the non-parametric statistics

Summarise the data

Summary of mesh sizes by type of net

Summary of depth of fishing by type of net

Summary of distance to shore by net

Summary of mesh sizes by craft type

Summary of depth of fishing by craft type

Summary of distance to shore by craft type

Plot to see the distribution of the data

Overall distribution via frequency curves/histograms and scatterplots

Plotting variables to visualise skew

Plotting box plots to show differences between fisheries and craft types

Plotting for bottom depth by fishery (net type)

Plotting for bottom depth by fishery (net type)

Plotting for bottom depth and distance of fishing by craft type

Plotting for mesh sizes across fisheries and craft type

Kruskal Wallis for non parametric

Differences in between gear, craft & regions

post hoc with FDR correction - Benjamini-Hochberg adjustment

SPATIAL ANALYSIS

Spatial patterns

Prepare the data set for neighbourhood analysis

Global spatial correlation:local geary's c

Spatial patterns for mesh size distributions

Prepare the data

Run the Geary test

Spatial autocorrelation test on categorical data:join counts

Getis-Ord G-test

plotting distance violations

Mechanised craft

Mesh size violations

Generate plots

generate histogram of distance from shoreline

distance of all craft

distance per craft type

Depth per craft type

Processing graphs and data on violations

Mapping

Get the basemaps and panels

Stamen map

Plot of distances

Plot panels of different variables

GIS data processing using GRASS and R

Create a 10km grid over the data and clean up legacy files

Create subsets of maps

run for fishery

run for boat type

run for class of craft

Subset map based on mesh size

Process craft classes

Build quadrats around fisheries

Build quadrats around mesh sizes

Generate interactive maps

Craft and mesh size

Mesh classes and their distribution

Net types and mesh sizes

Craft classes and mesh sizes